


* Models tests not included.  Ask authors

* estimate final models with dichotomous urban/rural variable

logit  pre_el nonurban nonpart retention appoint contested noton POL_SOPH reside age black hispanic other female married pid judinterest [pweight = V102] 


dog
* Estimate final post model with dichotomous urban/rural variable

logit post_el nonurban nonpart retention appoint contested noton POL_SOPH reside age black hispanic other female married pid judinterest [pweight = V102] 


Dog

* estimate percent correctly classified for pre- and post models


logit  pre_el nonurban nonpart retention appoint contested noton POL_SOPH reside age black hispanic other female married pid judinterest [pweight = V102] 

pred

dog

logit post_el nonurban nonpart retention appoint contested noton POL_SOPH reside age black hispanic other female married pid judinterest [pweight = V102] 

pred

dog

* estimate predicted probabilities for key variables

logit  pre_el nonurban nonpart retention appoint contested noton POL_SOPH reside age black hispanic other female married pid judinterest [pweight = V102] 

prchange nonurban nonpart contested POL_SOPH reside pid judinterest, x(nonurban=1 nonpart=1 retention=0 appointj=0 contested=1 noton=0 black=0 hispanic=0 other=0 female=1 married=1)

dog

logit post_el nonurban nonpart retention appoint contested noton POL_SOPH reside age black hispanic other female married pid judinterest [pweight = V102] 

prchange nonurban nonpart contested POL_SOPH reside pid judinterest, x(nonurban=1 nonpart=1 retention=0 appointj=0 contested=1 noton=0 black=0 hispanic=0 other=0 female=1 married=1)

dog



* Perform analysis for population density using continuous measure

* Pre-analysis

logit  pre_el countypop nonpart retention appoint contested noton POL_SOPH reside age black hispanic ///
 other female married pid judinterest [pweight = V102] 
 

 * generate % correctly predicted
 
 pred
 
 Dog
 *create graph
 
mcp countypop, var1(20) show 
Dog

* Post-analysis

logit  post_el countypop nonpart retention appoint contested noton POL_SOPH reside age black hispanic ///
other female married pid judinterest [pweight = V102] 

Dog

* % correctly predicted

pred

* create graph
mcp countypop, var1(20) show 

*combine graphs

graph combine pop_pre.gph county_post.gph

* generate predicted probabilities for analysis using prgen

* Pre-analysis

logit  pre_el countypop nonpart retention appoint contested noton POL_SOPH reside age black hispanic ///
 other female married pid judinterest [pweight = V102] 
 
 prgen countypop, from(1901) to (9818605) ncases(50) gen(pre)

 
 * Post-analysis

logit  post_el countypop nonpart retention appoint contested noton POL_SOPH reside age black hispanic ///
other female married pid judinterest [pweight = V102] 

prgen countypop, from(1901) to (9818605) ncases(50) gen(post)

dog
